Research On The Method Of Food Safety Public Opinion Analysis Based On The Kernel Method

With the further development of the Internet,especially the mobile Internet,public access to information and opinions have become more convenient,and sensational events have become more frequent.As food safety concerns the national economy and people’s livelihood,food safety has always been the focus of social attention.The constant fermentation of topics can easily lead to major lyrical events and negatively affect society.This article studies the method of food safety public opinion analysis,discusses the data filling and topic detection and tracking algorithms in depth,and has important practical significance for ensuring the healthy development of the food industry and promoting social harmony and stability.In the public opinion analysis,data preprocessing is the basis of algorithm operation.In data preprocessing,the phenomenon of missing data is often encountered.In order to make the existing data fully applied,the algorithm will not cause deviation due to missing data.Missing data needs to be filled.In this paper,the KKMOD algorithm is proposed based on cluster analysis and outlier detection.Clustering analysis is used to distinguish the data types and ensure that only the data in the same cluster can influence each other and avoid other types of data interference.After that,the most similar clusters are selected.Fill in the data to improve its data to fill in the correct rate.Finally,it uses the outlier detection algorithm to iteratively correct the padding value,which further improves the data filling accuracy.Based on the data filling,this paper studies the topic detection and tracking algorithm.In order to obtain a suitable topic model,nuclear K-Means was used for topic detection and an initial topic model was constructed.In the topic tracking stage,in order to improve the phenomenon of topic drift caused by changes in the topic center of gravity,this paper conducts research on topic tracking based on SVM and improves it.First,join news-related attributes,participate in classification experiments,and calculate and compare credibility.Finally,the categories of the samples to be classified are determined.Secondly,the topics are adaptively modified so that the algorithm can automatically update the topic models,effectively track follow-up reports of topics,and improve topic drift.Finally,this paper designs and implements a food safety network public opinion analysis system.The system adopts topic crawlers to collect relevant data and fill in the missing data to form a complete food safety public opinion data set.Then a vector space model is constructed and the KKMOD proposed in this paper is used.The method and the improved SVM topic tracking algorithm can detect and track topics in the food safety public opinion data set,making the initial topic model construction and follow-up news tracking of the topic more accurate.